In a free-naming experiment, a participant smells an odor and must name it without choosing from a list. The request sounds simple. Yet the answer can be brief or circuitous, shared by several participants or unique to one person, given quickly or reached after a pause. It may name a supposed source — something burnt, a fruit, a cleaning product — rather than the quality of the smell itself.
In a broad comparison of twenty sampled languages, smells were relatively difficult to name on average, but the difficulty was neither uniform nor captured by a single measure. Some sampled speakers have produced more consistent and concise responses, and have more often used dedicated abstract odor terms, than matched English- or Dutch-speaking participants. This does not, by itself, tell us whose nose is more sensitive.
The puzzle becomes clearer once the apparently ordinary phrase naming a smell is taken apart. What counts as success: agreeing with other speakers, responding quickly, using a dedicated odor term, identifying the source the experimenter had in mind, or distinguishing one smell from another? Research has measured all of these outcomes, but they are not interchangeable.
What counts as naming well?
Suppose ten people encounter the same odor. If eight give the same label, their answers show high agreement. If all ten give different descriptions, agreement is low — even if each answer is meaningful to the person who gave it. This concentration of responses is often called codability: roughly, how consistently a group converges on a way of describing something under particular task conditions.
Researchers can also classify the form of an answer. A dedicated abstract odor term describes a smell quality without merely naming a likely source. A source-based response instead says what the smell resembles or might come from. Neither category is automatically “correct.” The distinction reveals how speakers package an experience in words.
Concision and response time add further information. A short, rapidly available term suggests a different naming pattern from a delayed chain of comparisons. But even these measures do not tell us whether a participant identified the source intended by an experimenter. And none is a direct test of perceptual acuity. A person can distinguish two odors without having tidy names for either of them, just as someone can recognize two faces while struggling to recall their names.
These distinctions matter because the claim that humans are “bad at naming smells” often fuses them into one imagined deficit. The cross-language evidence does not support such a simple diagnosis.
A counterexample to the universal story
In one matched comparison, Asifa Majid and Niclas Burenhult asked ten Jahai men and ten American English-speaking men to name twelve odors from the Brief Smell Identification Test, as well as color chips. The sampled Jahai speakers gave odor responses that were more concise and more consensual. The American English sample’s responses were longer, more varied and more often described possible sources.
A related pattern appeared in a later, larger study. Thirty Jahai participants and thirty matched Dutch participants named single-compound laboratory odors. The Jahai sample used dedicated abstract terms more often, produced shorter answers and responded faster.
Together, these results rule out an absolute claim that everybody performs poorly on the tested free-naming measures. They do not show that every Jahai speaker responds in the same way, nor that all English or Dutch speakers struggle. The studies also concern the same Jahai community and come from an overlapping research network. They are two related comparisons, not independent surveys of unrelated populations.
Most importantly, the measured difference was linguistic performance under particular tasks. It was not evidence that the Jahai participants had more sensitive noses, experienced more vivid smells or possessed a biologically superior sense.
That still leaves a problem. A striking counterexample can disprove a universal claim, but it cannot tell us how common the exception is. For that, the comparison has to widen.
The average survives the exception
A broader study analyzed 44,091 sensory descriptions from 313 respondents speaking twenty languages from fifteen language families. Participants described stimuli from several sensory domains, allowing researchers to compare how strongly speakers within each language converged on their responses.
There was no single sensory ranking shared by all twenty languages. The relative codability of domains varied, producing thirteen different rankings. Smell was nevertheless the least codable domain on average — but not in every language.
The two results therefore fit together. Across this diverse sample, naming smells was relatively difficult on average. Yet the average did not describe every sampled language, and the Jahai comparisons showed especially clear departures from the response patterns in the matched English and Dutch samples. A tendency can be genuine without being a human universal.
The twenty-language study was broad, not exhaustive. Several languages came from the same families, and languages and cultures did not form twenty statistically independent cases. In other words, the dataset cannot be treated as twenty unrelated draws from humanity. Nor was it designed to estimate how all the world’s languages would rank the senses. It gives us evidence of substantial variation and an average disadvantage for smell within that dataset — not a global league table.
What an experiment allows us to see
If naming performance varies, it is tempting to treat every score as a transparent reading of a fixed ability. But experimental success also depends on what researchers ask, record and count.
Two experiments with Dutch speakers illustrate the point. Odors with common source labels were sometimes identified more accurately, but the association depended on the experiment and on how responses were scored. It appeared in the second experiment. In the first, it appeared when only the initial response was counted, but not when all responses were included. Label frequency also could not be cleanly separated there from the age at which a word had been learned.
A separate manipulation presented odors either in random order or in blocks grouped by whether their source objects were edible or inedible. It did not produce the predicted semantic-interference effect.
The lesson is narrower — and more useful — than “common words solve the problem.” Label properties matter, and the scoring rule can change whether an association appears in the analysis. These Dutch experiments do not explain why the Jahai and Dutch samples differed.
The way linguistic resources are collected matters too. In research on Huehuetla Tepehua, broad initial elicitation had revealed almost no olfactory ideophones: vivid conventional forms associated with sensory qualities. Targeted work using several methods uncovered forty-five. Separate documentation of Cha’palaa found abstract smell roots, a grammatical construction used to talk about odors and odor language in recorded discourse.
These findings show that conventional resources for discussing smells can remain invisible when the method is too general. They also show that odor language lives outside a single laboratory prompt. But a documented inventory is not an experimental performance score. Finding forty-five forms does not establish how often every speaker uses them, how quickly people would produce them in free naming, or whether they identify odors more accurately.
Method affects what becomes visible; it does not make the cross-language variation unreal.
Why no single cause wins
Once the variation is established, an obvious question follows: why do some sampled groups converge more readily on odor terms?
Lexical convention is one plausible part of the answer. If a speech community has dedicated terms that are routinely used, those forms may be more readily available during a naming task. The documented grammars, ideophones and discourse practices show that such conventions exist. Yet the direction of influence remains uncertain. Regularly talking about odors might support a more elaborated vocabulary, an established vocabulary might make such talk easier, or both could develop together.
Ecology, subsistence and everyday practice have also been proposed as contributors. Comparative work has reported associations involving hunter-gatherer communities, and a study of two closely related communities with different subsistence patterns sharpened that possibility. But subsistence is not something researchers can randomly assign. In that comparison, it remained bundled with differences in language, cultural practices, attitudes and experience. The evidence is observational, so the attractive sentence “this way of life causes better smell naming” goes beyond what the studies can show.
Cross-cultural variation does not rule out general processing constraints on odor naming. An independent review argued that olfactory object naming faces particular neurocognitive constraints. Such constraints could help explain the average difficulty while leaving room for linguistic convention, practice and cross-community flexibility. Task design, meanwhile, clearly affects measured outcomes but does not by itself explain the differences between communities.
The strongest positive cross-cultural findings also come largely from one collaborative research program and recurring communities. That does not erase the reported results. It does limit how confidently they can be turned into a single mechanism, a worldwide prevalence estimate or a causal story.
Naming is not smelling
The most seductive mistake is to translate a difference in words into a difference in sensory worlds. More agreement, shorter answers or a larger documented vocabulary can sound like evidence of finer perception. The central experiments did not test that proposition.
In the Jahai–Dutch study, verbal responses differed while the groups’ initial facial responses to pleasant and unpleasant odors were similar. That comparison concerns initial valence rather than fine discrimination, but it illustrates why verbal and nonverbal outcomes must be kept apart.
An independent experiment published in 2023 tested another part of the boundary. Participants in one group were told the odors’ identities before they had to discriminate among odor mixtures. In that particular task, the labels did not alter discrimination. This is not proof that language can never influence any aspect of perception. It is evidence that supplying labels and distinguishing mixtures are separable performances, and that a verbal difference cannot simply be promoted into a perceptual one.
The same caution works in the other direction. Saying that naming is not perception does not make naming trivial. A shared abstract term is a real linguistic resource. Agreement among speakers is a real feature of communication. The cross-language differences are worth explaining precisely because they reveal variation in how experiences become publicly nameable — without ranking the experiences themselves.
Return, then, to the participant facing an odor and an unrestricted request for a name. A hesitant, indirect answer may reflect the terms that are conventionally available, their familiarity, the way the question was posed and what the experiment counts as success. It does not offer a transparent reading of the nose.
Explaining why these patterns vary will require more independent, comparable studies that measure verbal naming and nonverbal perception without treating one as a substitute for the other.
The task has not changed. What has changed is what we think a hesitant or indirect answer can tell us.
Sources
- Majid, Asifa, and Niclas Burenhult. “Odors are expressible in language, as long as you speak the right language.” Cognition, 2014.
- de Valk, Josje M., et al. “Olfactory language and abstraction across cultures.” Philosophical Transactions of the Royal Society B, 2018.
- Majid, Asifa, et al. “Differential coding of perception in the world’s languages.” Proceedings of the National Academy of Sciences, 2018.
- Huisman, John L. A., and Asifa Majid. “Psycholinguistic variables matter in odor naming.” Memory & Cognition, 2018.
- Floyd, Simeon, Lila San Roque, and Asifa Majid. “Smell Is Coded in Grammar and Frequent in Discourse: Cha’palaa Olfactory Language in Cross-Linguistic Perspective.” Journal of Linguistic Anthropology, 2018.
- O’Meara, Carolyn, Susan S. Kung, and Asifa Majid. “The Challenge of Olfactory Ideophones: Reconsidering Ineffability from the Totonac-Tepehua Perspective.” Current Anthropology, 2019.
- Olofsson, Jonas K., and Jay A. Gottfried. “The muted sense: neurocognitive limitations of olfactory language.” Trends in Cognitive Sciences, 2015.
- Majid, Asifa, and Nicole Kruspe. “Hunter-Gatherer Olfaction Is Special.” Current Biology, 2018.
- Majid, Asifa. “Human olfaction at the intersection of language, culture, and biology.” Trends in Cognitive Sciences, 2021.
- Cormiea, Sarah, and Jason Fischer. “Odor discrimination is immune to the effects of verbal labels.” Scientific Reports, 2023.